A number of findings indicate that we tend to take seriously the evidence that supports what we believe and ignore the evidence against it. The resulting persistence of belief has been implicated in psychological disorders, educational difficulties, faulty decision-making, and political and social conflict. The irrationality of such persistence can be established by showing that we violate two principles: 1., neutral evidence should not strengthen belief; 2., we should not be affected by the order of evidence when order is truly irrelevant. The polarization effect (Pitz, 1969; Lord, Ross, & Lepper, 1979), in which neutral or mixed evidence causes people of different initial opinions to polarize, shows that the first principle is violated. The primacy effect (Hovland, 1957) suggests that the latter is as well. In both cases, however, artifactual accounts need to be tested. Several (non-competing) accounts of these phenomena will be tested, some of which involve cognitive biases such as bolstering, confusion of (moral and empirical) arguments, or the use of a single mental model corresponding to the favored belief. Others involve general beliefs about the conduct of thinking, and specific desires that certain beliefs be true. These hypotheses will be tested in several experiments involving polarization (and, in some cases, primacy). These experiments will involve comparisons of different kinds of content, attempts to evoke different mechanisms directly, and studies of individual differences. Another project will involve tutorial interviews of different types, designed to correct possibly irrationally perseverent beliefs. Another will follow up earlier findings concerning biases in the search for evidence. A final project will begin a longitudinal study of the relationship between self-critical thinking in planning and the long-term success of plans. The results should inform us about the conditions under which irrational persistence is likely to occur, and how it might be corrected or prevented.